fshy

Home / Hugging Face

Hugging Face public Gradio chat Spaces (anonymous API)

huggingface.co/docs/hub/spaces-zerogpuInference through public Gradio Spaces, anonymous callersinference/huggingface-gradio-chat-spacesJSONcomparereport an errorprovider JSON

Free allowance
ZeroGPU Spaces give an unauthenticated caller 2 minutes of GPU quota a day at low queue priority (5 minutes for a free account); a public Gradio chat Space publishes its API schema with no token
Verdict
caveatusable, with a named limit or an unprobed read path
Cost
free tier
Identity
anonymousno card, no phone
Binding limit
volume-quota2 minutes of ZeroGPU GPU quota per day for an unauthenticated caller, low queue priority; 5 minutes for a free account; the quota resets exactly 24 hours after the first GPU usage
Evidence
documented4 sources; a first-party page was read and the sentence quoted
Freshness
stale on 2026-11-29checked 2026-10-29 (live); stale on 2026-11-29
Free tierZeroGPU Spaces give an unauthenticated caller 2 minutes of GPU quota a day at low queue priority (5 minutes for a free account); a public Gradio chat Space publishes its API schema with no token
unit minutes; metered on wall-clock; each Space defines its own endpoint names, such as /generate on huggingface-projects/llama-2-13b-chat; there is no OpenAI-shaped route and no model id echo
Pricenone stated
Limits
  • volume-quota (binding) 2 minutes of ZeroGPU GPU quota per day for an unauthenticated caller, low queue priority; 5 minutes for a free account; the quota resets exactly 24 hours after the first GPU usage source
Bindingvolume-quota the 2 minutes a day of GPU quota binds; a Space on CPU hardware has no such quota stated in the page read
Identityanonymous an unauthenticated caller is a documented quota tier (Unauthenticated, 2 minutes, Low queue priority); no token was needed to read /gradio_api/info
Egressnot published
Interfacerest
Toolinggradio_client or the Space's /gradio_api/call/<api_name> routes; the schema is at /gradio_api/info on current Gradio and /info on older ones
Composes withpairs with ad hoc scripts; the 2 minutes a day of ZeroGPU quota for an unauthenticated caller is the scarce link
Reliabilitya Space can be paused or in an error state, and an older Gradio Space serves its schema at /info rather than /gradio_api/info
Evidence notedocumented the quota is first-party; the schema was read by a keyless GET. No chat call was made because probe issues GET and HEAD only, so a completed answer is unverified
Parsed numbers
  • volume-quota 2 minutes
Evidence 4
datesourcewhat it proved
2026-10-28huggingface.co/docs/hub/spaces-zerogpuan unauthenticated caller has 2 minutes of daily ZeroGPU quota at low priority
2026-10-28huggingface.co/docs/hub/spaces-zerogpua free account has 5 minutes a day at medium priority
2026-10-28huggingface.co/docs/hub/spaces-zerogputhe daily quota resets 24 hours after first use
2026-10-28huggingface-projects-llama-2-13b-chat.hf.space/gradio_api/ina keyless GET /gradio_api/info answered 200 with a named /generate endpoint on a public chat Space
Unknown values 3
  • $.egress.metered egress metering was not published
  • $.per_object_cap.stated the field does not apply to this kind of offer
  • $.retention.stated the field does not apply to this kind of offer